The short answer is no thank you.
It would not be wise to do it the 'unsafe' way, especially when there is a high voltage kick. As I mentioned, the attenuator needs to be grounded to do its job and protect the scope and allow you to scale the high voltage.
When you have the scope ground on a voltage supply, you cannot ground the scope.
Therefore it would be impossible to show the safe method on two channels and the unsafe method on the other two since the scope must be grounded.
Frankly, I would not want to look at anything with a high voltage kick and have the scope ground attached to the device rather than ground. This puts the scope and the PC at risk of high voltage kick.
You can get away with doing that with low voltage stuff using one channel only. One example might be placing signal and ground for one channel across a floating ground VRS crank sensor, such as on a Ford EDIS system, so you can measure true peak to peak amplitude. With math channels though, it's really not necessary.
Bottom line.....Unless you are really confident in what you are doing, never place the scope ground on a voltage source.
(The Picotologist) http://www.autonerdz.com
THE PicoScope Automotive Authority
In North America